arXiv:0706.3187v1 [astro-ph] 13 Jun 2007 Simulations of the Poynting–Robertson Cosmic Battery in Resistive Accretion Disks Dimitris M. Christodoulou, 1 Ioannis Contopoulos, 2 and Demosthenes Kazanas 3 ABSTRACT We describe the results of numerical “2.5–dimensional” MHD simulations of an initially unmagnetized disk model orbiting a central point–mass and respond- ing to the continual generation of poloidal magnetic field due to a secular source that emulates the Poynting–Robertson (PR) drag on electrons in the vicinity of a luminous stellar or compact accreting object. The fluid in the disk and in the surrounding hotter atmosphere has finite electrical conductivity and allows for the magnetic field to diffuse freely out of the areas where it is generated, while at the same time, the differential rotation of the disk twists the poloidal field and quickly induces a substantial toroidal–field component. The secular PR term has dual purpose in these simulations as the source of the magnetic field and the trig- ger of a magnetorotational instability (MRI) in the disk. The MRI is especially mild and does not destroy the disk because a small amount of resistivity damp- ens the instability efficiently. In simulations with moderate resistivities (diffusion timescales up to ∼16 local dynamical times) and after ∼100 orbits, the MRI has managed to transfer outward substantial amounts of angular momentum and the inner edge of the disk, along with azimuthal magnetic flux, has flowed toward the central point–mass where a new, magnetized, nuclear disk has formed. The toroidal field in this nuclear disk is amplified by differential rotation and it can- not be contained; when it approaches equipartition, it unwinds vertically and produces episodic jet–like outflows. The poloidal field in the inner region cannot diffuse back out if the characteristic diffusion time is of the order of or larger than the dynamical time; it continues to grow linearly in time undisturbed and without saturation, as the outer sections of many poloidal loops are being drawn radially outward by the outflowing matter of high specific angular momentum. On the other hand, in simulations with low resistivities (diffusion timescales larger than 1 Math Methods, 54 Middlesex Turnpike, Bedford, MA 01730. E-mail: [email protected]2 Research Center for Astronomy, Academy of Athens, GR–11527 Athens, Greece. Email: [email protected]3 NASA/GSFC, Code 663, Greenbelt, MD 20771. E-mail: [email protected]
81
Embed
Simulations of the Poynting-Robertson Cosmic Battery in Resistive Accretion Disks
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
arX
iv:0
706.
3187
v1 [
astr
o-ph
] 1
3 Ju
n 20
07
Simulations of the Poynting–Robertson Cosmic Battery
in Resistive Accretion Disks
Dimitris M. Christodoulou,1 Ioannis Contopoulos,2 and Demosthenes Kazanas3
ABSTRACT
We describe the results of numerical “2.5–dimensional” MHD simulations of
an initially unmagnetized disk model orbiting a central point–mass and respond-
ing to the continual generation of poloidal magnetic field due to a secular source
that emulates the Poynting–Robertson (PR) drag on electrons in the vicinity of
a luminous stellar or compact accreting object. The fluid in the disk and in the
surrounding hotter atmosphere has finite electrical conductivity and allows for
the magnetic field to diffuse freely out of the areas where it is generated, while at
the same time, the differential rotation of the disk twists the poloidal field and
quickly induces a substantial toroidal–field component. The secular PR term has
dual purpose in these simulations as the source of the magnetic field and the trig-
ger of a magnetorotational instability (MRI) in the disk. The MRI is especially
mild and does not destroy the disk because a small amount of resistivity damp-
ens the instability efficiently. In simulations with moderate resistivities (diffusion
timescales up to ∼16 local dynamical times) and after ∼100 orbits, the MRI has
managed to transfer outward substantial amounts of angular momentum and the
inner edge of the disk, along with azimuthal magnetic flux, has flowed toward
the central point–mass where a new, magnetized, nuclear disk has formed. The
toroidal field in this nuclear disk is amplified by differential rotation and it can-
not be contained; when it approaches equipartition, it unwinds vertically and
produces episodic jet–like outflows. The poloidal field in the inner region cannot
diffuse back out if the characteristic diffusion time is of the order of or larger than
the dynamical time; it continues to grow linearly in time undisturbed and without
saturation, as the outer sections of many poloidal loops are being drawn radially
outward by the outflowing matter of high specific angular momentum. On the
other hand, in simulations with low resistivities (diffusion timescales larger than
field diffusion (compare Figs. 3c, 4c) and this allows the torus to survive for many subsequent
orbits.
The inflowing matter builds another torus in the vicinity of the central point–mass
(Fig. 5). This nuclear torus continues to receive inflowing matter and azimuthal flux from
the inner edge of the original torus. The main characteristics near the equatorial pressure
maximum of the nuclear torus are summarized in Table 1, where large values of the ratio
BZ/Bφ signify that an axial outflow has already occurred and the toroidal field has been
released from the nuclear torus. For κ = 0.1, the diffusivity is not sufficiently strong to drive
the magnetic flux out of the nuclear region and no significant wind–like outflow is observed
in the vicinity of the central object (the diffusion “bubbles” seen in Fig. 5c between the two
tori are quite weak and devoid of azimuthal flux, as Fig. 5b indicates). When the nuclear
azimuthal flux becomes just a few percent of the equipartition value, it is released vertically
in a collimated jet outflow (Figs. 5 and 6) reminiscent of the “plasma–gun” mechanism
described by Contopoulos (1995). Such outflows are usually bipolar, but occasionally they
are asymmetric when the field is expelled preferentially in one or the other direction (Figs. 8
and 10). The asymmetry develops on the equator of the nuclear torus because the interface
between toroidal fields with opposite polarities is unstable (it is fluttering up and down). The
instability is seen in Figs. 5b and 9b, where this interface is curved throughout the region
between the central point–mass and the inner edge of the original torus. On occasion, the
nuclear toroidal field becomes weaker when it is expelled from the center and then we can
see the complex structure of the much weaker field in and around the original torus (Figs. 7b
and 9b).
– 16 –
The poloidal magnetic flux
Ψ(R, 0) ≡ 2π
∫ R
0
BZ(R′)R′dR′ , (31)
is monitored over the entire equatorial plane of the computational grid (Fig. 11). Because
the PR source generates complete poloidal loops, the total poloidal flux over the equator of
the grid is zero (loops crossing the equator in one direction have to turn back eventually and
cross in the opposite direction).
The conservation of Ψ(Rmax, 0), where Rmax is the radial edge of the grid, is broken
after 18 orbits when magnetic field from the surface of the torus reaches the outer edge
of the grid and outflows. Then poloidal loops open up at R = Rmax (Figs. 6c, 7c) and
Ψ(Rmax, 0) becomes permanently positive (Fig. 11). In the nuclear torus, the enclosed flux
switches polarity several times and also undergoes high–frequency oscillations due to the
episodic evolution of the nuclear magnetic field (see panels c in Figs. 5–10). At all the other
equatorial radii, and especially in the regions between the two tori and within the original
fluid, the poloidal flux increases linearly with time and this increase continues for more than
80 orbits (Fig. 11), when large amounts of mass and angular momentum reach the radial edge
of the computational grid. Such a steady, gradual increase of the poloidal field was predicted
by CK and CKC for moderate and high levels of diffusivity and owes its linear character to
the time independence of the PR source term S that was included in the induction equation
(eq. [8]).
4.2. Low Resistivity Models With κ ≤ 0.01
In models with κ ≤ 0.01, the diffusion of the field is strong enough to dampen the MRI
and to delay the organized inflow of matter for at least 10 orbits (e.g., Figs. 12a and 13a).
Slow diffusion and the fluttering instability cause the toroidal field to spread away from the
surface of the fluid where it is created. Some of this field is ejected into the surrounding
atmosphere but another part of it diffuses into the fluid of the torus (Fig. 12b). Eventually,
the inner edge of the torus is destabilized by the MRI, loses part of its angular momentum,
and an unbroken stream (a ”sheet”) of inflowing matter is created that is threaded by strong
toroidal magnetic field (Figs. 12b, 14b). In the low resistivity models, a nuclear disk is not
formed. Instead, lumps of matter with embedded field that reach the center ahead of the
inflowing stream are ejected vertically in asymmetric outflows (e.g., Fig. 13a, b).
The entire evolution of these models is reminiscent of the corresponding ideal MHD
model (κ = 0; Fig. 15) except that the ideal MHD fluid is dominated by strong oblique
– 17 –
shocks that distort the torus (Fig. 15a) and that are not observed in models with κ ∼ 0.01
(Fig. 14a). We note that the lump of fluid seen at the center in Fig. 15a will not be the
seed for the formation of a nuclear disk; it will soon be ejected vertically and it will clear
the center for more lumps to come in ahead of the organized inflowing stream that needs
another 6 or 7 orbits to get into the same area.
The poloidal magnetic flux at different equatorial radii of the model with κ = 0.01 is
shown in Fig. 16. Although the flux is initially growing linearly with time, this growth is
quickly terminated after about 23 orbits. This is because the toroidal field remains nearly
frozen into the fluid and does not unwind (e.g., Fig. 14b, c). Eventually magnetic reconnec-
tion and the fluttering instability along the equator of the grid limit field growth, while the
new field generated by the PR source is not significant in magnitude to make a difference.
This behavior is intimately linked to the inability of the low–resistivity models to form a
nuclear disk. We have determined by additional simulations that all models with κ ≤ 0.06
present the same characteristics, while two models with κ = 0.065 and κ = 0.07 exhibit
nuclear–disk formation within 3 − 4 orbits and uninterrupted central flux amplification for
20 orbits when these runs were terminated.
4.3. High Resistivity Models With κ ≥ 1
In models with κ ≥ 1, field diffusion occurs over dynamical timescales and the field
quickly spreads out to the surrounding atmosphere and inward to the fluid of the original
torus. The MRI is damped very efficiently in these simulations and the original torus survives
for more than 140 orbits (Figs. 17–20) when large amounts of outflowing matter and angular
momentum have crossed the outer radial edge of the computational grid. For the first few
orbits, the dominant field is toroidal and it is being built up continually by the differential
rotation of the fluid in the original torus. Most of this field is carried into the nuclear region
where it becomes anchored in the newly formed nuclear torus and diffuses away from it in all
directions (Figs. 17b, 18b and 20b). Inside the original torus, the field is limited efficiently
by magnetic reconnection; it remains very weak, and this is why its does not appear in the
contour plots of the same figures.
The nuclear torus continues to receive the inflowing matter and azimuthal flux. The
main characteristics near its equatorial pressure maximum are summarized in Table 2. After
the first 30 orbits, this torus ends up rotating faster than the nuclear torus of the standard
model. The fluid is cooler and the axial outflows are weaker and appear much later (at
τ ∼ 100). Magnetic pressure support remains always at a level of at least ∼ 1% of the fluid
pressure, and as a result, the nuclear fluid is always less dense than that of the standard
– 18 –
Table 2:Nuclear Disk in the High Resistivity Model With κ = 1
τ ρ/ρmax Pfl β Bφ BZ/Bφ vφ vZ/vφ
11.01 0.3 5.2 × 10−10 114 1.1 × 10−6 9.8 0.06 3.7
37.66 48.7 3.2 × 10−9 88 1.8 × 10−5 1.4 0.7 0.9
100.11 6.7 1.1 × 10−8 49 6.2 × 10−5 0.7 0.9 1.0
141.91 3.8 6.4 × 10−9 180 4.1 × 10−6 7.2 0.1 2.3
Notes.—Plasma β ≡ Pfl/Pmag, ρmax ≡ 10−10.
model.
Interestingly, in models with κ ∼ 10, there is no sign of inflowing matter or the de-
velopment of the MRI for more than 10 − 15 orbits, which implies that the instability is
suppressed efficiently by field diffusion. But in models with κ ∼ 1, matter and azimuthal
flux inflow does occur over short timescales and the evolution proceeds initially as in the
standard model with a nuclear torus forming in 2− 4 orbits. But the magnetic field diffuses
easily out of this nuclear structure and this process weakens the development and the input
power of the jets in all the models with κ ≥ 1. For κ = 1, some mild jets (vZ ∼ 1 − 2 and
BZ ∼ Bφ) are finally observed after about 100 orbits (Fig. 19b, c).
The equatorial poloidal flux grows again linearly in time at all radii outside of the
nuclear torus, just as in the standard model. This is shown in Fig. 21 for the model with
κ = 1. The conservation of Ψ(Rmax, 0) = 0 is now broken sooner because of the higher rates
of diffusivity. Fig. 21 shows that after just 8 orbits the field loops reach the outer edge of
the grid and open up (see also Fig. 17c); then Ψ(Rmax, 0) becomes permanently positive.
Furthermore, the poloidal flux within the nuclear torus remains positive at all times (which
indicates the presence of a large–scale organized poloidal field; panels c in Figs. 18–20) and
undergoes again high–frequency oscillations due to radial fluctuations of the material to
which the poloidal–field lines are attached.
5. Summary and Discussion
In this work, we have performed detailed numerical, multidimensional, MHD simulations
of the Poynting–Robertson (PR) battery, a mechanism that is capable of generating cosmic
magnetic fields in the vicinity of luminous, accreting, compact and stellar objects. The PR
effect was included in the simulations of an accretion–disk model orbiting a central point–
– 19 –
mass by introducing a continuous source of poloidal magnetic field into the induction equation
(eq. [8]). At the same time, the differential rotation of the disk model provided an elemental
source of toroidal magnetic field by twisting dynamically the poloidal field lines. The fluid
in the accretion disk and in a surrounding tenuous nonrotating atmosphere was resistive
and allowed for the magnetic field to diffuse away from the areas where it was originally
produced. In all models, a large–scale accretion flow was established from the initial disk
model toward the central point–mass by the action of a magnetorotational instability (MRI)
in the orbiting fluid. Two of the above features, the PR current that acts as a continuous
source of weak magnetic field and the global accretion flow that may cause its amplification
by drawing field of a single polarity to the center, are what sets the PR battery apart from
previously proposed and critically reviewed mechanisms of field generation and amplification
such as the Biermann (1950) battery and the turbulent dynamo process (see also § 1).
The present simulations constitute a first attempt toward studying the global magne-
tohydrodynamics of resistive large–scale accretion flows and the possible amplification or
saturation of the generated magnetic flux in the presence of various degrees of magnetic
dissipation. The latter is controlled by a free parameter, the resistive frequency κ (eq. [30])
which is a direct measure of the resistivity η of the fluid and inversely proportional to the
electrical conductivity σ. We have found that a value of κ ≈ 0.06 (corresponding to a diffu-
sion timescale τdiff ≈ 16 local dynamical times) is the critical value that separates two types
of physically different model evolutions:
1. Models with moderate and high resistivities (κ > 0.06) exhibit strong field amplifica-
tion that continues uninterrupted for over 100 orbits (Figs. 11 and 21). In about 2− 4
orbits, the inflowing matter creates a nuclear torus near the central point–mass and
the magnetic field that is transported into the nucleus by accretion and by diffusion
becomes anchored onto this torus. When the nuclear toroidal field becomes strong, it
unwinds and produces episodic bipolar jet–like outflows, in addition to the diffusing
field bubbles that are observed to emerge from the center when κ ≥ 1. The equato-
rial field is unstable to fluttering and this instability is responsible for the occasional
appearance of markedly asymmetric vertical jets and for the ejection of magnetic field
into the surrounding atmosphere. All of these details are illustrated in Figs. 2–10 for
our standard model with κ = 0.1 and in Figs. 17–20 for the κ = 1 model.
2. Models with low–resistivities (κ ≤ 0.06) exhibit some moderate field amplification for
about 20 orbits, but then the magnetic field quickly saturates to dynamically insignif-
icant levels because of the weak diffusion and the absence of unwinding of the toroidal
component, as the magnetic field remains nearly frozen into the matter. The accretion
flow carries its magnetic field toward the central point–mass but it does not create a
– 20 –
nuclear torus. Eventually magnetic reconnection, the fluttering instability, and some
asymmetric ejections of magnetized lumps limit the growth of the field to less than 3
orders of magnitude above the values seen early in the model evolutions (Fig. 16). All
of these details are illustrated in Figs. 12–14 for the κ = 0.01 model and in Fig. 15 for
the ideal MHD model with κ = 0.
The above results are in agreement with those discussed by CK and CKC on the basis
of qualitative arguments and more idealized model calculations. The present work provides
further evidence in support of our original conclusions and we are confident that the proposed
battery mechanism will prove important to the theory of generation of cosmic magnetic fields.
The critical value of the inverse magnetic Prandtl number determined by CK, namely
(Pm)−1 ≃ 2, also appears to be in agreement with the critical value of κ ≈ 0.06 determined
from the present simulations, if an allowance is made for a rough, order-of-magnitude es-
timate of the effective viscous timescale τvis associated with turbulent, MRI–driven inflow
from the initial torus under ideal–MHD conditions (when the dynamics is not altered by
resistive slipping of the magnetic field through the matter): In our κ = 0 simulation with
frozen–in magnetic field, the inflowing stream of matter has not reached the center after 19
orbits (Fig. 15) and the continuing evolution shows that this is still the case after 26 orbits
when the stream is getting close to the center. Based on this observation, we estimate that
τvis ≈ 26τdyn, where τdyn is the local dynamical time in the initial torus. Since the critical
κ ≈ 0.06 implies that τdiff ≈ 16τdyn, then the critical inverse magnetic Prandtl number in
the resistive simulations is
(Pm)−1 =τvis
τdiff
≃ 1.6 , (32)
and field amplification occurs for κ > 0.06 or, equivalently, for (Pm)−1 > 1.6. We note that
the above value of τvis implies also an effective value of αmag>∼ 0.04 for the analogue of the
Shakura–Sunyaev (1973) parameter of the accretion flow initiated by the MRI in the ideal–
MHD model. This is just a rough estimate and as such it is not out of line compared to values
determined previously from simulations of the MRI in the ideal–MHD limit (αmag ∼ 0.1;
Hawley & Krolik [2002] and references therein). But notice that the effective αmag–parameter
increases dramatically to a value of αmag ≈ 0.3−0.5 in the κ > 0.06 models in which a robust
nuclear disk forms in just a few orbits. We have to conclude then that a sufficient amount
of magnetic diffusivity appears to be the cause of dynamical nuclear–disk formation in the
above resistive models.2
2This conclusion should not be confused with the conclusions of Stone & Pringle (2001) and Hawley &
Balbus (2002) who see the fluttering instability and the formation of the nuclear disk but find no purely axial,
collimated outflow and no substantial differences in their models when a small amount of artificial resistivity
– 21 –
All the simulations of models with κ > 0 show that field diffusion works against the
MRI and this instability is damped with increasing success as the value of κ is increased.
This result is known and well–understood (e.g., Fleming, Stone, & Hawley 2000; Fleming
& Stone 2003). When the magnetic field is allowed to slip through the matter, then the
field lines cannot hold on to specific fluid elements and facilitate their exchange of angular
momentum and azimuthal magnetic flux. However, the MRI is not eliminated from any
model with a reasonable value of κ and the weakened modes continue to transport some
of the angular momentum to larger radii and matter with enhanced azimuthal magnetic
flux toward the central point–mass (see also Christodoulou, Contopoulos, & Kazanas 1996,
2003). In the moderate and high–resistivity models (κ ≥ 0.1), the transfer of these conserved
quantities is gradual and this allows the original accretion tori to survive for hundreds of
orbits (in our simulations, it takes 80 − 140 orbits for substantial amounts of matter and
angular momentum to cross the outer radial edge of the grid, a distance only twice as large
as the characteristic size of the initial torus).
In our model evolutions, we strengthened artificially the PR source because the current
state of computing does not allow us to run MHD models with a weak PR source and wait
for millions of dynamical times to see whether the magnetic field will be amplified or not
(§ 2.2). Even with an artificially enhanced PR source, however, the initial magnetic field is
1 − 2 orders of magnitude smaller than that utilized to induce an MRI in previous MHD
simulations (e.g. Hawley 2000; Hawley, Balbus, & Stone 2001): At early times (τ >∼ 0.1), the
poloidal field grows at the inner edge of the initial torus to ∼ 10−9, a value that results in a
plasma β ∼ 5 × 104. Within the first orbit, the toroidal field also catches up in magnitude
and at later times, all field components are amplified as the magnetic field is drawn into the
nuclear region. The amplification of the poloidal flux is eventually interrupted (at τ ∼ 20) in
the low–resistivity models, as discussed above and in § 4.2. The amplification of the toroidal
flux in the κ > 0.06 models is also interrupted episodically by the repeated unwinding of the
toroidal field that accumulates in the nuclear torus. This effect is a version of the “plasma–
is included to smear out current sheets. These simulations were essentially carried out using ideal MHD;
as such, they can see the intrinsic instability of the equatorial toroidal field and the large–scale structure
of the accreted fluid, but they cannot capture the influence of moderate or large amounts of anomalous
resistivity. Furthermore, the inflowing stream in these ideal–MHD simulations reaches the center in just 2
orbits. Such inflow is too fast, essentially dynamical, and it does not appear in line with the values of the
effective αmag–parameter (∼ 0.05− 0.1) reported for the fluid in the original torus when it is destabilized by
the MRI. Stone & Pringle (2001) claim that global stresses of the radial magnetic field are responsible for
this early outward transport of angular momentum and the associated inflow that occurs before the MRI
actually becomes nonlinear. But no such global stresses are observed in our κ = 0 simulation in which the
magnetosonic rarefaction waves take time to transverse the fluid, to redistribute the conserved quantities
locally, and finally to drive the MRI into the nonlinear regime.
– 22 –
gun” expulsion suggested by Contopoulos (1995): when the toroidal field grows close to
equipartition in the nuclear torus, it can no longer be confined; it is released dynamically
in the vertical direction and its stresses act to confine the poloidal–field component close to
the symmetry axis of the torus. This situation is evident in many of the diagrams (panels
c) shown in § 4.1 and § 4.3, where the poloidal field lines are nearly vertical at small radii
and the “funnels” along the Z–axis are extremely narrow. At the same time, the b–panels
of the same figures depict the substantial degree of collimation imposed by the toroidal field
to the low–density matter flowing within these funnels. These results are of interest because
they demonstrate that the anomalous resistivity in the accretion flows and the plasma–gun
mechanism may be responsible for producing the highly collimated jets observed in a variety
of accretion–powered galactic and extragalactic objects (see, e.g., Bridle & Perley 1984;
Mirabel & Rodrıguez 1999; and Wilson, Young, & Shopbell 2001).
Furthermore, our results indicate that the magnetic diffusivity plays a more important
role in accretion disks than previously thought. For moderate or large values of this pa-
rameter (or, equivalently, for values of the resistive frequency κ > 0.06), there is a clear
tendency in the models to generate and maintain strong, well–ordered, large–scale poloidal
magnetic fields which couple to the rotation of the nuclear flow and result in matter expulsion
along the rotation axis. In contrast, no such features are seen at large scales in models with
κ ≤ 0.06. Therefore, our models suggest that the well–known and well–defined dichotomy of
accretion–powered objects (e.g., Xu, Livio, & Baum 1999; Ivezic et al. 2004) to those ex-
hibiting powerful jets (“radio–loud”) and those lacking such structures (“radio–quiet”) may
be related to and should be sought in the physics that determines the value of this particular
macroscopic parameter of the accretion flows that power the emission of these objects. The
calculations presented here are only a preliminary step toward exploring this notion.
This work was supported in part by a Chandra grant.
REFERENCES
Balbus, S. A., & Hawley, J. F. 1991, ApJ, 376, 214
Balbus, S. A., & Hawley, J. F. 1998, Rev. Mod. Phys., 70, 1
Biermann, L. 1950, Z. Naturforsch, 5a, 65
Bisnovatyi-Kogan, G. S., & Blinnikov, S. I. 1977, a, 59, 111
Bisnovatyi-Kogan, G. S., Lovelace, R. V. E., & Belinski, V. A. 2002, ApJ, 580, 380
– 23 –
Bridle, A. H., & Perley, R. A. 1984, ARA&A, 22, 319
Christodoulou, D. M., Cazes, J. E., & Tohline, J. E. 1997, New Astronomy, 2, 1
Christodoulou, D. M., Contopoulos, J., & Kazanas, D. 1996, ApJ, 462, 865
Christodoulou, D. M., Contopoulos, J., & Kazanas, D. 2003, ApJ, 586, 372
Christodoulou, D. M., & Sarazin, C. L. 1996, ApJ, 463, 80
Contopoulos, J. 1995, ApJ, 450, 616
Contopoulos, I., & Kazanas, D. 1998, ApJ, 508, 859 (CK)
Contopoulos, I., Kazanas, D., & Christodoulou, D. M. 2006, ApJ, 652, 1451 (CKC)
Evans, C. R., & Hawley, J. F. 1988, ApJ, 332, 659
Fleming, T. P., & Stone, J. M. 2003, ApJ, 585, 908
Fleming, T. P., Stone, J. M., & Hawley, J. F. 2000, ApJ, 530, 464
Hawley, J. F. 2000, ApJ, 528, 462
Hawley, J. F., & Balbus, S. A. 2002, ApJ, 573, 738
Hawley, J. F., Balbus, S. A., & Stone, J. M. 2001, ApJ, 554, L49
Hawley, J. F., & Krolik, J. H. 2002, ApJ, 566, 164
Igumenshchev, I. V., Narayan, R., & Abramowicz, M. A. 2003, ApJ, 592, 1042
Ivezic, Z., Richards, G. T., Hall, P. B., Lupton, R. H., Jagoda, A. S., Knapp, G. R., Gunn,
J. E., Strauss, M. A., Schlegel, D., Steinhardt, W., & Siverd, R. J. 2004, ASP Conf.
Ser., 311, 347
Kulsrud, R. M., Cen, R., Ostriker, J. P., & Ryu, D. 1997, ApJ, 480, 481
Mirabel, I. F., & Rodrıguez, L. F. 1999, ARA&A, 37, 409
Narayan, R., & Yi, I. 1994, ApJ, 428, L13
Papaloizou, J. C. B., & Pringle, J. E. 1984, MNRAS, 208, 721
Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337
Stone, J. M., Hawley, J. F., Evans, C. R., & Norman, M. L. 1992, ApJ, 388, 415
– 24 –
Stone, J. M., & Norman, M. L. 1992a, ApJS, 80, 753
Stone, J. M., & Norman, M. L. 1992b, ApJS, 80, 791
Stone, J. M., & Pringle, J. E. 2001, MNRAS, 322, 461
Tohline, J. E. 1988, unpublished
Vainshtein, S. I., & Rosner, R. 1991, ApJ, 376, 199
van Leer, B. 1977, J. Comp. Phys., 23, 276
van Leer, B. 1979, J. Comp. Phys., 32, 101
Wilson, A. S., Young, A. J., & Shopbell, P. L. 2001, ApJ, 547, 740
Xu, C., Livio, M., & Baum, S. 1999, AJ, 118, 1169
This preprint was prepared with the AAS LATEX macros v5.2.
– 25 –
FIGURE CAPTIONS
Fig. 1.— Flowchart for the completion of one timestep in the MHD code.
Fig. 2.— Standard model with κ = 0.1 at time τ = 0.76. The first rarefaction wave
propagates outward in the fluid and a substantial toroidal field with amplitude Bφ = 6.1 ×10−7 has been built by differential rotation on the surface of the torus. Panel a: Mass density
contours as solid lines, angular momentum density contours as dashed lines, and poloidal
momentum densities as vectors. Panel b: Contours of the toroidal magnetic field (solid
lines in the +φ direction and dashed lines in the −φ direction) and vectors of the poloidal
magnetic field. In each of these two panels, contours are drawn down to the 5% level of the
corresponding maximum value and arrows are drawn down to 10% of the largest magnitude.
Also, very small vectors with magnitudes between 1% and 10% of the maximum value are
replaced by dots in order to indicate in which regions of the grid the vector fields tend to
spread. Panel c: Poloidal–field lines irrespective of field magnitude; the detailed structure
of the very weak field can be seen here as well. The data are padded with zeroes at R = 0
in order to delineate the behavior of the field lines near the Z–axis.
Fig. 3.— As in Fig. 2 but for τ = 1.19. The inner edge of the torus is destabilized by the
MRI.
Fig. 4.— As in Fig. 2 but for τ = 1.93. Matter and field have flowed into the nuclear region
and an episodic vertical jet–like outflow has developed (|vZ| ≈ 18 at the base of the jet, as
opposed to vφ ≈ 5.4). The unwinding of the nuclear toroidal field results in a substantial
axial field near the Z–axis (see Table 1).
Fig. 5.— As in Fig. 2 but for τ = 4.59. High–angular momentum fluid has flowed to larger
radii away from the outer edge of the torus, another (nuclear) torus has formed near the
central point mass by inflowing matter, while a prominent jet has developed in the vertical
direction away from the nucleus (|vZ | ∼ 1 − 2 at the base of the jet).
Fig. 6.— As in Fig. 2 but for τ = 20.85. The nuclear torus has become denser than the
original torus (ρ = 6.6×10−10), while the jet–like outflow appears to be very well collimated
and bipolar.
Fig. 7.— As in Fig. 2 but for τ = 29.61. The original torus has flattened substantially due
to inflowing and outflowing matter, while the fluttering instability has expelled the toroidal
field from the nuclear torus that appears to be weakly magnetized (Bφ = 2.5×10−7; see also
Table 1). Panel b then shows the structure of the relatively weak magnetic field (Bφ ∼ 10−6)
– 26 –
that has spread into the original torus and in the surrounding atmosphere.
Fig. 8.— As in Fig. 2 but for τ = 59.72. The original torus has separated into two regions
and the outer region is moving outward. The nuclear torus has become very dense, hot, and
strongly magnetized (Table 1). This torus appears to also support an asymmetric jet–like
outflow with a strong magnetic field (Pmag ∼ Pfl) embedded into the diffuse (ρ ∼ 10−12)
outflowing matter.
Fig. 9.— As in Fig. 2 but for τ = 92.46. The original torus continues to feed matter to the
nuclear region and to move radially outward, while another vertical outflow (|vZ | ∼ 2 at its
base) is taking place in the nuclear torus.
Fig. 10.— As in Fig. 2 but for τ = 110.93. The nuclear disk has expelled much of its own
angular momentum in a wind and it has also developed another asymmetric vertical outflow.
Only 9.7% of the initial mass and 2.0% of the initial angular momentum remain within the
computational grid at this time.
Fig. 11.— Poloidal magnetic flux Ψ(R, 0) on the equatorial plane of the grid, integrated
out to different radii, for the standard model with κ = 0.1. The integrated flux beyond the
nuclear torus increases linearly with time for over 80 orbits, while the flux within the nuclear
torus oscillates at very high frequencies and switches polarity several times.
Fig. 12.— Low resistivity model with κ = 0.01 at time τ = 7.50. Two rarefaction waves carry
angular momentum outward in the torus, a ”sheet” of inflowing matter has developed toward
the nucleus, and the fluttering instability has pushed field into the surrounding atmosphere.
The toroidal field presents quite a complex distribution, but it is weak in magnitude (∼ 10−7
or smaller).
Fig. 13.— As in Fig. 12 but for τ = 10.38. Matter in the nuclear region does not get
organized in a disk; instead it is ejected asymmetrically in the vertical direction (vZ ≈ 5)
along with its embedded toroidal field (Bφ ∼ 10−6), while a weak axial–field component
(BZ ∼ 10−7) also develops at small radii.
Fig. 14.— As in Fig. 12 but for τ = 16.30. No nuclear disk has developed and the asymmetric
vertical outflow continues in the inner region, while the inflowing sheet of material appears
to be threaded by strong magnetic field (all components are ∼ 10−6). Only 3.5% of the
initial mass and 3.7% of the initial angular momentum has flowed out of the computational
grid at this time.
Fig. 15.— Ideal MHD model with κ = 0 at time τ = 18.86. No field diffusion occurs in
this model and the field remains permanently frozen into the fluid. This model evolution
– 27 –
is similar to the low resistivity model shown in Fig. 14 above, except for the strong oblique
shocks observed here within the fluid of the original torus and at the tip of the inflowing
sheet. Only 3.1% of the initial mass and 3.2% of the initial angular momentum has flowed
out of the computational grid at this time.
Fig. 16.— Poloidal magnetic flux Ψ(R, 0) on the equatorial plane of the grid, integrated out
to different radii, for the low resistivity model with κ = 0.01. The integrated flux no longer
increases linearly with time at times τ > 23.
Fig. 17.— High resistivity model with κ = 1 at time τ = 11.01. A nuclear torus has formed
from inflow and a strong magnetic field (BZ ∼ 10−5, Bφ ∼ 10−6) is anchored onto it (see also
Table 2). Two field ”bubbles” have expanded out of the center and have diffused obliquely
into the surrounding atmosphere.
Fig. 18.— As in Fig. 17 but for τ = 37.66. The nuclear torus has become very dense
(ρ ≈ 5× 10−9) and a vertical outflow has developed (|vZ| ∼ 1 at its base) in addition to the
obliquely expanding bubbles.
Fig. 19.— As in Fig. 17 but for τ = 100.11. The original torus has been flattened by the
MRI, while the strongest field (BZ ≈ 8 × 10−5) participates in a collimated jet–like outflow
(|vZ| ∼ 2 at its base) anchored at the nuclear torus.
Fig. 20.— As in Fig. 17 but for τ = 141.91. The original torus has spread toward the outer
edge of the grid as rarefaction waves continue to redistribute angular momentum, the nuclear
torus has transported outward much of its own angular momentum in a wind, the fluttering
interface instability has disrupted the sheet of inflowing matter, and a vertical jet is seen
along with magnetic–field bubbles that diffuse obliquely out of the center. Only 8.6% of
the initial mass and 7.0% of the initial angular momentum remain within the computational
grid at this time.
Fig. 21.— Poloidal magnetic flux Ψ(R, 0) on the equatorial plane of the grid, integrated
out to different radii, for the high resistivity model with κ = 1. As in the standard model
(κ = 0.1 in Fig. 11), the integrated flux beyond the nuclear torus increases linearly with time
for over 140 orbits, while the flux within the nuclear torus oscillates at very high frequencies